The Mathematical Modelling of Rugby

While the use of machine learning as a tool within statistical analysis has become standard in sports such as football, rugby has been slower to adopt these techniques due to the complexity of the sport. However, analytics using these tools can be applied to many different parts of the game, ranging from individual player analysis to team performance. This thesis aims to create the basis for machine learning applications within rugby by developing models that measure key aspects of the game, including attacking actions and decision-making. In collaboration with the IRFU, this research begins by developing an expected threat (xT) statistic similar to that used in football, before presenting a model which calculates what decision should be made at a penalty. These models serve as a foundation for future machine learning in rugby, with potential applications in player evaluation and tactical analysis.